Re: problem using netboot and root nfs

[Torsten Fischer <tof@bee.de>]

Martin Dudle wrote:
>
> hi there
>
> i'm stuck with the following problem:
>
> using netboot 0.6.1 to boot a 2.0.30 kernel via network. root filesystem
> gets mounted by nfs. the whole thing actually works to the point
> after root filesystem was mounted by the client. i took a look
> to the kernel sources. there i saw that (since execute_command is NULL)
> it tries to execve() /etc/init, then (if it fails) /bin/init, then
> /sbin/init. /sbin/init should be executed since it's on the root fs.
> unfortunately it isn't. i don't get any error messages, too.
> next in the kernel comes a while(1) loop that tries to execute
> /bin/sh (using kernel_thread()). but not shell ever gets started.
> kernel_thread() seems to work since it returns pids that increase and
> make sense.

Hello,

it could be that the init-prgram isn't executed it is in the search
path. This happens if the init-program is dynamically linked and the
client bootet for the first time. Usually the init-program executes the
program ldconfig to setup the environment for the ld-prgram. This is
required by all dynamically linked programs. Without running ldconfig at
least one time, ld won't work. Without ld, all dynmically linked
programs won't work. If your init is dynamically linked, it can't start.
So your client is in an endless loop.

There are two possibilities to get around this problem. First , to use a
static init program. Second, make a chroot into the root-directory of
your client and execute a ldconfig by hand. After that, your client
should startup.

Hope this helps.

Regards,
	Torsten